This is an open access article distributed under the terms of the Modern collective action increasingly relies on networked media to connect, mobilize, and sustain transnational participation in protests. Yet the digital platforms and tools that make this possible can also undermine solidarity by widening gaps in how protests are experienced and perceived between on-the-ground and online participants, and between those at home and those abroad. Diaspora members, for example, physically distanced from events in the field, often rely on the internet and secondhand accounts, whereas residents experience protests firsthand and face more direct risks. Such disparities in participatory patterns, distance, and risk exposure may result in varying protest experiences and mutual misperceptions, particularly in how people feel during protests and how they make sense of protest actions. Existing research has explored the political activism of diaspora communities and their relations with home (Keck & Sikkink, 1998; Moss, 2018; Ogan, 2020; Sheffer, 2003), as well as the role of online campaigns and social media in protest participation (González-Bailón & Wang, 2016; Jost et al., 2018). However, less attention has been paid to how participation mode and distance may affect resident and diaspora activists’ protest representations and their perceptions of each other’s experiences.
This study examines how resident and diaspora members represent protests and predict each other’s experiences, focusing on both the emotional tone of protests (“how protests feel”) and how protest actions are construed (“why” vs. “how”). Using Iran’s 2009 post-election dissent movement as a case study, I address three questions: (1) What emotions characterize protest experiences among residents and diaspora members, and how accurately do they predict each other’s positive and negative affect during protests? (2) How do the two groups construe different protest actions, and how closely do they predict each other’s emphasis on the “how” (mechanics) versus “why” (purposes)? (3) Do the type and intensity of activism predict these representations and (mis)perceptions?
The goal is to isolate where cross-border activism breaks down: not only in what each group reports, but also in what they assume about the other side. Residents may question the motives and legitimacy of diaspora activism and doubt whether diaspora members understand what protest actions mean and why they are taken, whereas diaspora activists, relying on mediated information, may misread how protests feel on the ground and extrapolate from the most salient episodes. Such misperceptions can erode trust and coordination and weaken transnational coalitions. Pinpointing where perceptions diverge can yield concrete targets for strengthening collaboration and sustaining movement cohesion. The results can also inform how policymakers and outside observers assess diaspora members’ roles in transnational protests and interpret protest information across sources.
Diaspora Activism, Digital Networks, and Cross-Border Perceptions
Diaspora communities can exert significant influence on the society and culture of their home countries (Anderson, 1998; Basch et al., 2005; Moss, 2020; Newland, 2010), and at times have helped drive major sociopolitical movements from abroad (Bermudez, 2011; Gertheiss, 2015; Smith & Bakker, 2011). Historically, scholars have attributed this capacity to the civil and political freedoms granted by host countries, as well as sociocultural developments that inspire diaspora members to advocate for change in their homeland (Lacroix et al., 2016; Vertovec & Cohen, 1999; Waldinger & Fitzgerald, 2004). In practice, diaspora repertoires include material support (e.g., remittances, humanitarian goods, professional expertise), agenda-setting and voice amplification abroad, and political organizing (e.g., lobbying and movement leadership).
More recently, online connectivity and the proliferation of social media have expanded the toolkit available to diaspora communities, enabling them to build extensive transnational networks, coordinate participation, disseminate information, and shape public opinion in their home and host countries (Aziz, 2024; Caren et al., 2020; Kok & Rogers, 2017). Social networks can accelerate mobilization, influence participation rates, and shape perceptions of protest size and impact (Barberá et al., 2015; Greijdanus et al., 2020; Hassanpour, 2014). As diaspora activism becomes increasingly intertwined with these digital affordances, research on the role of online networks during protests can shed light on home–diaspora relations in transnational collective action.
Reviewing evidence from protest movements in the United States, Spain, Turkey, and Ukraine, Jost et al. (2018) identified two broad functions of online social networks in protests: the dissemination of emotional themes and motivational appeals, and the facilitation of information exchange. Since diaspora activists are often not subject to the same restrictions as residents, they can react more quickly, speak more freely, and reach wider audiences (Earl et al., 2022; Feldstein, 2022). Emotional themes can help construct a shared reality of protest successes, setbacks, and repression, reinforcing the moral justification for continued collective action. This motivational/moral frame is paired with an operational one, focused on logistical and informational support to organize and coordinate resident protesters. By relaying and amplifying voices from within across borders, diaspora members facilitate communication among domestic actors, which can further contribute to mobilization and coordination (González-Bailón & Wang, 2016; Tofangsazi, 2023). In this sense, these functions map onto the “why” and the “what/how” of protests, and alignment on both is likely critical for sustaining collective action.
At the same time, these functions can also introduce systematic distortions in how events are perceived across the field–online divide and across resident–diaspora lines. Emotional themes and motivational appeals may bias how diaspora members perceive on-the-ground events. Such appeals often mobilize participation through moral outrage, social identification, and group efficacy (Jost et al., 2018), processes closely aligned with core motives for collective action (van Zomeren, 2013). For example, Langer et al. (Langer et al., 2019) analyzed 80,000 tweets linked to the Occupy Wall Street movement and found that 67% contained emotional expressions, 91% appealed to a shared or collective sense of efficacy, and 88% referenced concerns about fairness, social justice, or deprivation. Similarly, (Brady et al., 2021) showed that online moral outrage can be amplified by social feedback and norm learning within networks, intensifying emotional reactions and driving engagement in protests. Reliance on these mediated streams may therefore lead diaspora members to engage selectively with content that reinforces perceptions of protests as more intense and distressing than residents themselves report.
Online participation may also shape how diaspora activism is perceived by residents. Diaspora activists, leveraging freer access to the internet and perhaps compensating for their physical absence, may concentrate on sharing logistical and tactical information about the protests, including coordinating turnouts across locations, tracking counterprotest activity, and sharing tutorials on tools and workarounds. From the perspective of action identification theory (Vallacher & Wegner, 1989), such content emphasizes low-level “how” construals over higher-level “why” construals that foreground motives, goals, and meaning. For example, locking a door can be construed as “securing a house” (why) or “turning a key in the lock” (how). An emphasis on low-level details may foster a sense of efficacy within diaspora networks, but it may also lead resident activists to perceive them as out of step with the reasons certain actions are taken and how they advance the movement’s goals. What can seem helpful from afar—such as sharing information about the location of a new protest—could be seen domestically as a risky invitation misaligned with local resources and objectives, contributing to perceptions of diaspora activists as disconnected from lived realities.
In sum, mediated activism may impair diaspora members’ ability to gauge the valence and intensity of residents’ experiences, while making them seem more focused on execution than on the movement’s purposes and local constraints. These gaps can weaken cross-border coalitions. The present study tests these ideas by comparing how residents and diaspora members represent Iran’s 2009–2012 post-election protests and how they predict each other’s experiences.
Context: Iran’s 2009–2012 Post-Election Protests
In the wake of the 2009 Iranian presidential election and the incumbent’s re-election to a second term, a wave of demonstrations erupted in Tehran and other cities across Iran. These protests, dubbed the Green Movement, continued sporadically until early 2012, overlapping with the Arab Spring uprisings in the Middle East (Dabashi, 2017; Filin, 2022). They were among the first large-scale political upheavals revealing the power of the internet and social media as mobilization tools (Morozov, 2009; Pew Research Center, 2009). Online spaces provided anonymity, diversified modes of participation, and were less constrained than traditional media. These same features, however, made them vulnerable to misinformation, surveillance, and free riding, which undermined the credibility of digital content. As arrests and crackdowns stalled the protests, the cost differential between local and remote participation became more salient. In turn, internal critique gained momentum within the movement, particularly between residents and diaspora members whose engagement was largely mediated online.
Preliminary qualitative data from interviews, focus groups, and online content analysis shed light on how residents and diaspora members perceived and evaluated each other. Several themes emerged among residents critical of diaspora members’ online activism. Some argued that migration removed diaspora members from the realities of daily life in Iran and protest conditions, leaving them as biased spectators with outdated views at best. Others contended that regardless of commitment to the cause, diaspora members lacked the legitimacy to advise residents or propose solutions. A recurring theme was that a grasp of the status quo is necessary but not sufficient for evaluating protest tactics and strategies because insulation from the costs of failed action—violence, arrest, or confinement—would distort judgment. In this view, divergent risk profiles meant that one should not “talk the talk” unless one can also “walk the walk.” A smaller set voiced broader distrust of remote activism, casting certain efforts as disingenuous or as a pastime for those too removed to bear responsibility for the actions they encourage.
By contrast, diaspora activists maintained that they retained strong ties to Iran. They questioned whether distance or leaving Iran had diminished their cultural affinity and attachment or eroded their intuitive grasp of its inner workings. Some noted that they left precisely because of deep involvement in protests and civic or political activism. In their view, they were no less attuned to or concerned about Iran’s affairs than many residents and had incurred substantial personal and professional costs. Others emphasized that activism from abroad is not risk-free: migration and exile bring their own challenges and vulnerabilities, especially when family and friends remain in Iran. Still others argued that the “resident–diaspora” (or “insider–outsider”) divide was overstated, potentially amplified by actors seeking to fragment the movement by driving a wedge along identity and political fault lines.
Overall, both groups raised plausible arguments, suggesting that resident–diaspora tensions may stem from genuine differences in protest experiences as well as systematic misperceptions of each other’s realities. During Iran’s post-election protests, diaspora activists were less present on the streets but played prominent roles online by amplifying residents’ struggles to outside audiences, sustaining motivation and solidarity, and sharing tactical guidance (e.g., first-aid information, coordination messages, and updates on crowd movements and security forces; (Carafano, 2009; Michaelsen, 2018). By contrast, residents were more likely to attend street demonstrations, exposing them to greater personal risk and more direct encounters with repression. Such differences in participation mode and distance may produce distinct information ecologies: diaspora engagement is mediated through emotional appeals and online information exchange, whereas residents’ experience is anchored in on-the-ground participation and local constraints. This asymmetry can contribute to divergent protest representations.
At the same time, the divide may reflect misperceptions rather than actual differences in protest experiences. Even if residents and diaspora members hold broadly similar representations of protest experiences, they may misjudge what the other group feels during protests and how it construes protest actions. If so, we would expect prediction gaps to vary by activism and information access, with individuals more involved in protest-related activities showing smaller errors in cross-group inferences. In this view, resident–diaspora tensions may arise less from differences in experience than from systematic prediction errors about the other group’s experiences, shaped by differences in participation mode and by how online platforms structure remote engagement.
Study Overview
To test these possibilities, I compare Iranian residents’ and diaspora members’ (a) reports of protest experience—emotional tone (“how it feels”) and action construal (“why” vs. “how”), and (b) their predictions of each other’s reports during Iran’s 2009–2012 post-election movement. I ask whether the two groups (1) differ in their own reports, (2) their predictions of each other, and (3) whether both reports and prediction errors vary with activism.
Participants completed two tasks either from their own perspective or by predicting the other group’s responses. Task 1 assessed affective representations of protest, asking participants to picture themselves taking part in the protests and rank a set of words by how salient they were and how closely they matched what they would see and feel. Task 2 examined construals of protest-related activities, distinguishing higher-level “why” construals (motives/meaning) from lower-level “how” construals (execution/mechanics). For each activity, participants chose which description fit better (e.g., “expressing dissent” vs. “chanting slogans”). I then examine whether participants’ own representations and cross-group prediction errors covary with different forms and levels of activism, as well as diaspora participants’ degree of contact with Iran.
Method
Participants
The sample was predominantly male, urban, and highly educated, with a mean age in the early 30s. Residents and diaspora participants were broadly similar in age and urban residence, but the latter included a higher proportion of women and postgraduate degrees and was less likely to identify as Muslim. Diaspora participants also reported substantial ongoing contact with Iran, with many having visited recently and nearly half traveling at least annually. Details are available in Tables S1 and S2 of the Supplementary Appendix (SA).
Measures
Place of Residence and Condition Assignment
Participants first reported whether they currently resided in Iran (yes/no), which classified them as residents versus diaspora and determined assignment to the corresponding conditions in the protest-representation tasks.
Types and Levels of Activism
All participants then responded to questions about their activism and protest participation:
Protest attendance: ‘Which of the post-election protests did you attend?’ Responses were, ‘None’, The first week after the election’, ‘The first month after the election’, ‘The summer after the election’, ‘The first year after the election’, ‘First year and after’ (coded 1-6).
Specific protest activities: ‘How often have you engaged in any of these activities?
Disseminating information online or otherwise about the protests.
Taking to the streets in participation in public collective action.
Gathering by an Iranian embassy in protest.
The responses were, ‘Never’, ‘Once’, ‘More than once’, ‘Always’ (coded 1-4).
Overall time devoted to protest-related activities: ‘Overall, how much time have you dedicated to protest-related activities, online or offline, on a weekly basis?’ The responses were, ‘No activity’, ‘Less than an hour’, ‘1-5 hours’, ‘5-10 hours’, ‘10-20 hours’, ‘20-40 hours’, ‘More than 40 hours’ (coded 1-7).
Contact With Iran Among Diaspora Participants
Diaspora participants also responded to two additional questions: ‘How long has it been since your last trip to Iran?’ on a scale of, ‘About a month,’ ‘About 6 months,’ ‘About a year,’ ‘About 2 years,’ ‘About 3 years,’ and ‘More than 3 years’ (coded 1-6). The second question asked, ‘How often do you visit Iran?’ with response options: ‘Every 5 years or less,’ ‘Every 4 years or less,’ ‘Every 3 years or less,’ ‘Every 2 years or less,’ ‘Every year,’ ‘Twice a year,’ ‘Thrice a year or more,’ and ‘More than three times a year’ (coded 1-8).
Representations of Protest
Participants completed two tasks, either from their own perspective (Own condition) or from the perspective of the other group (Other condition). In the Other condition, residents responded as they thought diaspora members would respond, and diaspora participants responded as they thought residents inside Iran would respond. Task 1 focused on the emotional experience of protest, and Task 2 focused on how protest actions are construed. Note that because the instructions asked participants to “suppose you live inside/outside Iran and are answering as a resident/diaspora member,” responses likely reflect a mix of perspective taking and self-projection, rather than pure meta-perceptions about unrelated others in a strict sense.
Affective Experience in a Protest Scenario (Task 1)
All participants were asked to ‘think about the protest days, picture a day on the streets,’ which allowed inclusion of respondents without direct protest experience and kept the wording consistent across Own and Other conditions; random assignment should distribute prior protest experience across conditions. The complete prompt was as follows:
‘[From the perspective of residents/diaspora members], think about the protest days, picture a day on the streets. What do you[they] see and how do you[they] feel? Which of the following words are more present and salient in what you[they] perceive or sense? Which words are more vivid and palpable in this experience? Rank the following words by closeness, salience, and clarity in the protest that you[they] are picturing in your[their] mind.’
Stimuli were drawn from protest descriptions on social media and informal pre-survey interviews with residents and diaspora members (N = 15). Interviewees described what protest participation feels like and the images and sensations they associate with it. Frequently mentioned emotions that also appeared in social media content were selected. The final set comprised three eight-word lists (Table 1), each containing pairs of positive and negative emotion words, plus four neutral words (objects or bodily states). Word order within each list was randomized.
Table 1
Word Stimuli in the Affective Experience in a Protest Task (Task 1)
| Category | Set 1 | Set 2 | Set 3 |
|---|---|---|---|
| Negative Affect | grief | stress | doubt |
| fear | anger | powerless | |
| Positive Affect | excitement | capable | confidence |
| hope | harmony | joy | |
| Neutral | tear gas | blood | smoke |
| rock | baton | water | |
| fatigue | bruise | breathless | |
| nausea | burn | tear |
Construal Level of Protest Actions (Task 2)
Participants completed 12 Behavioral Identification Form (BIF) items (Vallacher & Wegner, 1989) shown in Table 2. After reading the original BIF instructions, they completed four practice items about everyday actions, followed by eight protest-specific actions presented in random order. The instructions were followed by:
‘Choose the identification (a or b) that best describes the behavior for you [residents/diaspora members]. Remember, mark the description that you believe [you think they believe] is more appropriate for each pair.’
Table 2
Behavioral Identification Form Items
| Action | High level (coded 1) | Low level (coded 0) | |
|---|---|---|---|
| a | paying rent* | Maintaining a place to live | Writing a check |
| b | eating* | Getting nutrition | Chewing and swallowing |
| c | reading a book* | Gaining knowledge | Following lines of print |
| d | locking the door* | Securing a place to live | Putting a key in the lock |
| 1 | voting | Influencing governance | Dropping a vote in the ballot box |
| 2 | resistance | Challenging authority | Saying ‘No’ |
| 3 | chanting | Showing solidarity | Shouting slogans |
| 4 | taking to the streets | Uniting forces for a cause | Gathering a crowd and marching |
| 5 | refrain from violence | Promoting peace | Avoiding injury or fighting |
| 6 | protesting | Defending one's rights | Taking part in street rallies |
| 7 | writing on walls | Publicizing grievance | Spraying on walls |
| 8 | covering one’s face | Concealing Identity | Wrapping one's face in a garment |
*are practice items from the original scale.
Demographics
The study concluded with demographic questions about sex, age, education, income, religion (Islam, no religion, other), and living area (urban, suburban, rural).
Procedure
The survey was first drafted in English to enable feedback from researchers and advisors who do not speak Persian. It was translated into Persian and reviewed by several native speakers to ensure clarity and cultural appropriateness; feedback was used to resolve ambiguities.
The survey began with an IRB-approved consent form outlining the study, the anonymity of responses, and the right to withdraw at any time. Participants who provided consent then reported their residence, followed by activism measures. Diaspora participants also answered two questions about contact with Iran. Next, participants were randomly assigned to complete the protest-representation tasks either from their own perspective (Own) or from the other group’s perspective (Other). Demographics were collected at the end.
Results
Protest representations are examined first, comparing residents’ and diaspora members’ own responses with their predictions of each other. Only complete responses are analyzed. I then test whether representations and prediction errors vary by type and level of activism.
Protest Representations
Affective Experience in a Protest Scenario (Task 1)
The rankings of positive (i.e., joy, excitement, confidence, hope, harmony, capable) and negative (i.e., grief, stress, doubt, fear, anger, powerless) affect words showed acceptable internal consistency (Cronbach’s α = .80 and .63, respectively)1. They were reverse-scored so that higher values indicate greater salience in protest representations and then averaged to create composite measures of positive (M = 5.38, SD = 1.68) and negative affect (M = 4.54, SD = 1.26). The two scores were significantly correlated (r = −.59, 95% CI [−.63, −.54], p < .0001). See the SA (Table S3) for descriptives and exploratory analyses of the neutral (non-affect) words.
Overall, participants in the Own condition ranked positive affect words higher than negative ones, indicating that positive emotions were more salient in representations of protests than negative emotions (paired Cohen’s d = −.32, 95% CI [−.46, −.17]2). Residents and diaspora members did not significantly differ in the extent to which they represented protests positively (Cohen’s d = −.06, 95% CI [−.26, .14]) or negatively (Cohen’s d = .08, 95% CI [−.13, .27]). However, the two groups diverged in how they predicted protest experiences from each other’s perspective.
Comparing predicted affect rankings in the Other condition with those from the Own condition, diaspora members predicted residents’ protest representations as significantly less positive (Cohen’s d = 0.21, 95% CI [0.032, 0.397]), and slightly more negative (Cohen's d = −0.19, 95% CI [−.37, −0.01]). In contrast, residents’ predictions of diaspora members’ experiences did not significantly differ from diaspora members’ own reports for either negative affect (Cohen's d = 0.044, 95% CI [−0.155, 0.242]) or positive affect (Cohen's d = 0.03, 95% CI [−.165, .222]). See Figure 1.
Figure 1
Positive (Panel A) and Negative (Panel B) Affect Ratings in Protests (Task 1)
Note. Panel A shows positive affect and Panel B shows negative affect. Dots represent participants’ mean ratings; bars show group means with 95% CIs. Lower plots show bootstrap estimates of unpaired Cohen’s d for predicted minus own ratings (left: Diaspora Predicted – Residents Own; right: Residents Predicted – Diaspora Own) based on 5,000 resamples; black dots indicate estimates, vertical lines 95% CIs, and shaded densities the bootstrap distributions.
Construal Level of Protest Actions (Task 2)
For the eight protest-related BIF items, low-level (“how”) responses were coded 0 and high-level (“why”) responses 1. Excluding resistance3, the items formed a scale with acceptable internal consistency (α = .6) and were averaged into an index for construal levels of protest actions. Action identification levels were generally high (M = .81, SD = .1, Mdn = .86).
In the Own condition, there were no significant differences between residents and diaspora members in the level at which they construed different actions (Cohen’s d = −0.052 95% CI [−0.208, 0.107]). In the Other condition, however, the two groups significantly differed (Cohen’s d = 0.25, 95% CI [0.088, 0.389]). Comparing predicted scores with those in the Own condition, residents expected lower levels of action identification among diaspora participants (Cohen’s d = −0.28, 95% CI [−0.44, −0.12]). In contrast, diaspora participants’ predictions did not differ significantly from residents’ Own scores (Cohen's d = −0.104 95% CI [−0.258, 0.054]). See Figure 2.
Figure 2
Protest Action Identification Ratings (Task 2) by Residence and Condition
Note. Plot elements follow Figure 1. Dots show participants’ mean action-identification ratings; bars show group means with 95% CIs; lower plots show bootstrap Cohen’s d (predicted − own).
Activism
Protest participation varied widely across activities (see SA, Table S6). Protest attendance and street presence were polarized: roughly two-thirds reported either no participation or sustained participation from the first year onward. Online information sharing was common, with 85% reporting doing so at least a few times. Embassy protests were rare (84% reported never attending) and were excluded from subsequent analyses. Weekly time commitment clustered at 1–5 hours (29%) and < 1 hour (20%), with 9% exceeding 20 hours.
Residents and diaspora members differed across activism measures (Figure 3). Compared to residents, diaspora participants reported lower protest attendance (Residents: M = 3.69, SD = 2.21; Diaspora: M = 2.83, SD = 1.96), t(1237) = 7.34, p < .001, d = −0.41, 95% CI [−0.52, −0.30], and less street participation (Residents: M = 2.60, SD = 1.16; Diaspora: M = 2.32, SD = 1.16), t(1197) = 4.32, p < .001, d = −0.25, 95% CI [−0.36, −0.13]). However, diaspora participants were more active online (M = 3.40, SD = 0.88) than residents (M = 3.20, SD = 1.03), t(1243) = −3.75, p < .001, d = 0.21, 95% CI [0.10, 0.32]. The two groups did not reliably differ in the time dedicated across activities (Residents: M = 3.10 (SD = 1.61); Diaspora members: M = 3.06 (SD = 1.53), t(1216) = 0.35, p = .72, d = −0.02, 95% CI [−0.14, 0.09]). See the SA for analyses controlling for demographics.
Figure 3
Standardized Activism Scores (z) Across Four Activity Measures by Residence Group
Note. Points represent individual responses (jittered for visibility); larger markers with error bars indicate group means with 95% confidence intervals.
Activism and Affective Experience in a Protest Scenario
To examine how activism relates to affective experience of protests, a linear mixed-effects model was tested with random intercepts for participants, predicting affect ratings with standardized activism scores entered in long format and moderated by activism measure (attendance, online information sharing, street protests, time dedicated), affect valence (Positive vs. Negative), residence (Iran vs. Abroad), and condition (Own vs. Other), controlling for demographics. The model revealed a robust activism × affect valence interaction, F(1, 5295.5) = 117.35, p < .001, such that higher activism was associated with more positive and less negative affect overall. By contrast, activism type showed little meaningful variability: there was no main effect of activism type, F(3, 5202.6) = 0.01, p = .998, and no activism × activism type interaction, F(3, 5792.4) = 0.10, p = .96. Only a small activism × activism type × affect valence, F(3, 5199.6) = 2.91, p = .033 emerged, with follow-up trend estimates indicating that activism slopes were similar across activities (approximately −0.18 to −0.05 for negative affect and 0.10 to 0.18 for positive affect).
Accordingly, the four standardized activism items were averaged into a composite activism index (α = .73; M = .00, SD = .74, range = −1.51 – 1.44), with higher scores indicating greater involvement in protests. Residents scored slightly higher on this index than diaspora participants (Cohen’s d = −0.15, 95% CI [−0.26, −0.04]). In the analyses that follow, this composite score is used to examine how activism relates to Own and Other affect ratings and to prediction accuracy.
An omnibus linear model predicting affect ratings with residence (Iran vs. Abroad), affect valence (Positive vs. Negative), condition (Own vs. Other), the composite activism index, and all interactions, with demographic covariates showed significant main effects of affect valence, F(1, 1497) = 28.74, p < .001, and activism, F(1, 1497) = 5.73, p = .017, qualified by a robust activism × affect valence interaction, F(1, 1497) = 54.19, p < .001. This interaction was further moderated by condition, activism × affect valence × condition: F(1, 1497) = 8.02, p = .005, and by condition and residence, activism × affect valence × condition × residence: F(1, 1497) = 5.58, p = .018. There was also a marginal affect valence × condition interaction, F(1, 1497) = 3.25, p = .072. Thus, the association between activism and affective ratings differed for Own versus Other conditions, and across residents and diaspora participants. Because the Own and Other conditions address conceptually distinct questions (how people represent protests vs. how they think the other group represents them), they were modeled separately (see SA, Table S4).
Own Condition
In the Own condition, long-form affect ratings were analyzed using a linear regression model, with the composite activism index as a predictor moderated by affect valence (Positive vs. Negative) and residence (Iran vs. Abroad), controlling for demographics. Results revealed a strong activism × valence interaction, F(1, 684) = 49.24, p < .001: higher activism was associated with more positive and less negative representations of protest. There was no significant three-way interaction with residence, activism × valence × residence: F(1, 684) = 1.45, p = .23, meaning residents and diaspora participants did not substantially differ in how activism influenced their affective experiences. Simple-slope estimates showed that higher activism predicted lower negative affect, b = −0.33, 95% CI [−0.57, −0.09], and higher positive affect, b = 0.71, 95% CI [0.48, 0.95], and the difference between these slopes was substantial, Δb = 1.04, SE = 0.17, t(684) = 6.23, p < .001. In particular, at the lowest levels of composite activism index (activity = −1.51), estimated negative affect was higher than estimated positive affect (M = 5.22 vs. M = 4.08), t(684) = 3.99, p < .001, Cohen’s d = 0.30, whereas at the highest levels (activity = 1.44), this pattern reversed, positive affect was estimated significantly higher than negative affect (M = 6.19 vs. M = 4.25), t(684) = 7.55, p < .001, Cohen’s d = 0.58.
Other Condition
A parallel model was fit for the affect ratings in the Other condition, revealing a reliable activism × valence interaction, F(1, 805) = 12.61, p < .001. Greater activism was associated with seeing the other group’s protest experience as relatively more positive (b = 0.33, 95% CI [0.13, 0.54]) and less negative (b = −0.20, 95% CI [−0.40, 0.01]). The difference between these slopes was also significant, Δb = 0.53, SE = 0.15, t(805) = 3.65, p < .001, mirroring the pattern in the Own condition but with smaller effect sizes. Unlike the Own condition, however, the activism × valence interaction was moderated by residence, activism × valence × residence: F(1, 805) = 4.95, p = .026, b = 0.64, 95% CI [0.08, 1.21], indicating that this pattern was more pronounced among diaspora participants than among residents.
Evaluating Predictions
Prediction errors for affect ratings were computed separately for positive and negative affect by subtracting the target group’s mean ratings in the Own condition from each participant’s prediction of that group. So for the diaspora participants, the negative-affect error score was their predicted mean rating for residents minus residents’ mean negative affect in the Own condition; for residents, it was their predicted mean for diaspora members minus diaspora members’ mean negative affect in the Own condition. The same procedure was applied to positive affect. Thus, positive error values indicate overestimation of the target group’s affect, and negative values indicate underestimation.
Direction of Predictions
A linear model predicting these signed error scores with activism, affect valence (positive vs. negative), residence (Iran vs. abroad), their interactions, and controlling for demographics, revealed a significant activism × valence interaction, F(1, 805) = 13.29, p < .001, which was further moderated by residence, activism × valence × residence: F(1, 805) = 4.95, p = .026. There was also a valence × residence interaction, F(1, 805) = 4.93, p = .027. Follow-up trend estimates showed that among diaspora participants, higher activism predicted more positive errors for positive affect, b = 0.52, 95% CI [0.20, 0.84], and more negative errors for negative affect, b = −0.33, 95% CI [−0.63, −0.02]. So the more active diaspora participants tended to describe residents’ protest experience as more positive and slightly less negative relative to residents’ own reports. Among residents, prediction errors were small and not reliably related to activism (negative affect: b = −0.06, 95% CI [−0.32, 0.20]; positive affect: b = 0.14, 95% CI [−0.11, 0.40]). The positive- and negative-affect slopes differed significantly for diaspora participants, Δb = −0.85, SE = 0.22, t(805) = 3.83, p = .0001, but not for residents, Δb = −0.21, SE = 0.19, t(805) = 1.11, p = .27. Figure 4 depicts these results.
Figure 4
Predicted Affect for the Other Group as a Function of Composite Activism and Residence
Note. Points show participants’ ratings in the Other condition (1–7). Columns show negative vs. positive affect; rows show resident vs. diaspora participants. Solid lines are fitted linear trends with 95% CIs. Dashed lines mark the target group’s mean ratings in the Own condition; vertical distance indicates prediction error. Diaspora participants’ predictions vary with activism (more positive, less negative), whereas residents’ slopes are near zero. Numbers in each panel represent standardized regression coefficients (β) and 95% confidence intervals from a linear mixed-effects model controlling for age, sex, education, income, area, and religion.
Accuracy of Predictions
To assess accuracy independent of direction, absolute errors were computed separately for positive and negative affect as the absolute difference between the participant’s prediction and the target group’s mean Own rating. These scores were modeled as a function of activism, affect valence, residence, and their interactions (with demographic covariates), revealing a main effect of valence, F(1, 805) = 48.92, p < .001: participants made larger absolute errors when judging positive than negative affect, b = 0.44, 95% CI [0.28, 0.60]. The omnibus test for the composite activism index was also statistically significant, F(1, 805) = 18.74, p < .001, but the corresponding regression coefficient was small and its confidence interval included zero (b = −0.05, 95% CI [−0.20, 0.09]), suggesting that this effect may reflect a more diffuse pattern across individual measures of activism rather than a strong, unitary signal.
Replacing the composite activism index with the four standardized activism measures entered simultaneously indicated that protest attendance and online information sharing emerged as unique predictors of smaller absolute errors. Absolute errors decreased with greater protest attendance (F(1, 805) = 17.67, p < .001, b = −0.09, 95% CI [−0.17, 0.01]), and greater online information sharing (F(1, 805) = 12.23, p < .001, b = −0.14, 95% CI [−0.21, −0.06]). Street protest and time dedicated were not reliably related to absolute error, Fs ≤ 2.40, ps ≥ .12. When protest attendance was excluded, street protest significantly predicted accuracy, suggesting that its apparent effect is largely captured by actual protest attendance.
Activism and Construal Levels of Protest Actions (Task 2)
A linear model was fit to logit-transformed action construal scores, predicted by activism scores entered in long format and moderated by activity type (protest attendance, online information sharing, street protest, time dedicated), residence (resident vs. diaspora participants), and condition (Own vs. Other), controlling for age, sex, education, income, area, and religion. There was a main effect of condition, F(1, 4387) = 49.11, p < .001 (higher-level identifications were more frequent in the Own than the Other condition), a main effect of residence, F(1, 4387) = 6.90, p = .009, and a residence × condition interaction, F(1, 4387) = 14.44, p < .001, such that the Own–Other difference was more pronounced for residents than for diaspora participants. The model also revealed a robust main effect of activism, F(1, 4387) = 20.91, p < .001: across activity types, participants higher in activism tended to construe protest behaviors at a higher level of identification. By contrast, activity type showed no meaningful variability: there was no main effect of activity type, F(3, 4387) < 0.01, p = .999, and no reliable activism × activity type interaction, F(3, 4387) = 1.51, p = .21. None of the higher-order interactions involving activity type reached significance (all Fs ≤ 1.51, all ps ≥ .21), justifying use of composite activism index.
Logit-transformed action construal score was entered as the response variable in a linear model, predicted by composite activism index, residence (resident vs. diaspora), condition (Own vs. Other), and their interactions, controlling for demographics. The model yielded a significant main effect of activism, F(1, 1232) = 12.12, p < .001: participants higher in activism were more likely to use higher-level action identifications. A main effect of condition, F(1, 1232) = 15.032, p < .001, moderated by residence, F(1, 1232) = 4.68, p = .030, showed that higher-level action identification was more frequent in the Own than in the Other condition. Follow-up contrasts showed that residents used more high-level identifications in the Own than in the Other condition (Own: M = 1.56, 95% CI [1.46, 1.67]; Other: M = 1.23, 95% CI [1.12, 1.34]; Δ = −0.33, SE = 0.08, p < .001), whereas the Own–Other difference was not significant among diaspora participants (Own: M = 1.55, 95% CI [1.43, 1.67]; Other: M = 1.47, 95% CI [1.35, 1.59]; Δ = −0.08, SE = 0.09, p = .35).
Separate linear models were estimated for the Own and Other conditions (see SA, Table S5). In the Own condition, the composite activism score showed a small but statistically detectable association with higher-level action identifications, F(1, 625) = 8.21, p = .004, b = 0.16, 95% CI [0.03, 0.28]. At the extremes of the composite activism index, participants with the lowest activism scores (-1.51) construed protest actions at significantly lower levels than those with the highest scores (1.44; M = 1.45 vs. M = 1.87), t(625) = 2.616, p < .0091, Cohen’s d = 0.21. Age was also positively associated with abstraction, b = 0.03, 95% CI [0.02, 0.03], p < .001; men relied less on higher-level identifications than women, b = −0.20, 95% CI [−0.38, −0.03], p = .025. In the Other condition, activism was a weaker predictor of action construal levels, F(1, 600) = 3.68, p = .056, b = 0.15, 95% CI [0.01, 0.31], and the activism × residence interaction was non-significant. However, residence showed a robust main effect, F(1, 600) = 8.20, p = .004, b = 0.23, 95% CI [0.06, 0.41], with diaspora participants attributing more high-level construals to the other group than resident participants did. Model comparison indicated that activism improved overall fit (ΔF(2, 596) = 3.46, p = .031), suggesting that it explained some unique variance even though the moderation by residence was weak.
Evaluating Predictions
Prediction error was computed as the participant’s predicted action-construal score (logit-transformed) minus the target group’s mean score in the Own condition (i.e., diaspora participants’ predicted residents score minus residents’ Own mean, and residents’ predicted diaspora score minus diaspora members’ Own mean). Because higher scores in this task indicate describing actions more in terms of their broader goals (“why”) rather than concrete steps (“how”), positive error values reflect overestimation of how the target group views protest actions, and negative error values reflect underestimation. Overall, errors showed a bias toward underestimation (M = −0.22, SD = 1.05).
Direction of Predictions
Prediction errors were regressed on the composite activism index, residence (resident vs. diaspora), and demographic covariates. The model explained a modest proportion of variance, R2 = .06, 95% CI [.02, .08]. There was a small but significant main effect of residence, F(1, 525) = 4.29, p = .039. After adjusting for activism and covariates, diaspora participants showed less negative bias (i.e., smaller underestimation) than residents, b = 0.19, 95% CI [0.01, 0.38], meaning that their predictions were closer to the target group’s own scores. That is, on average, their estimates were closer to how much residents construed protest actions in terms of broader goals rather than concrete means. Activism and the activism × residence interaction were not reliable predictors of signed error, all Fs ≤ 2.51, all ps ≥ .11.
Accuracy of Predictions
A similar model as above was fit with absolute error as the outcome to examine prediction accuracy independent of direction (R2 = .04, 95% CI [.00, .06]). Residence predicted accuracy, F(1, 525) = 9.18, p = .003: diaspora participants had smaller absolute errors than residents, b = −0.13, 95% CI [−0.24, −0.03]. Activism, however, did not reliably predict errors, F(1, 525) = 1.42, p = .23, b = −0.03, 95% CI [−0.12, 0.05], regardless of residence (activity × residence: F(1, 525) = 0.61, p = .43). When the four activism indices were entered simultaneously, protest attendance was the only measure that consistently predicted both the direction (b = 0.19, 95% CI [0.05, 0.32]) and magnitude (b = −0.09, 95% CI [−0.17, −0.02]) of construal-level errors. Higher attendance was associated with less underestimation of the target group’s construal levels and with smaller absolute errors.
Distance
Among diaspora participants, annual trips home and time since last visit were strongly correlated (r = −.65, t(558) = 20.07, p < .001, 95% CI [−.69, −.59]); they were recoded and averaged into a composite index of Distance (higher scores = less frequent trips and longer time since last visit; M = 4.03, SD = 1.87, Median = 4.50, range = 1.17 – 8). This score correlated with activism (r = −.22, 95% CI [−.30, −.14], p < .01), indicating that diaspora members with greater distance were also less engaged in protest activities. However, Distance was not significantly associated with participants’ own or predicted affective experiences, or with construal levels of protest actions (all |rs| < .08, ps > .30). Neither annual trips home nor time since last trip was associated with the direction or magnitude of affective prediction errors (|bs| ≤ 0.04, ps ≥ .51).
Discussion
In transnational protests, local and remote participants may misperceive each other’s experiences even when their own protest representations are broadly similar. Diaspora respondents in this study underestimated residents’ positive feelings (e.g., hope, confidence) and overestimated their distress, whereas residents saw diaspora activists as more focused on the execution of protest actions than on their broader goals. The extent to which these judgments reflect projection, perspective-taking, or both cannot be directly tested here. However, two findings are clear: (1) divergence lies less in people’s own reports than in their beliefs about the other side, and (2) these beliefs are linked to the type and level of activism.
While time devoted to activism was similar between the two groups, residents were more likely to attend street protests, whereas diaspora members reported greater online participation. This suggests that online versus offline participation may not fundamentally change people’s own protest representations, but it can shape how they perceive—and are perceived by—each other. Among participants with lower activism, remote participation was associated with more negative assumptions about residents’ on-the-ground experience, and with diaspora activists being perceived as more tactically (“how”) than strategically (“why”) oriented. Consistent with this, prediction errors were smaller among those who attended protests or shared information online, but overall time spent on activism was not reliably related to prediction accuracy.
Misperceptions about affective experiences during protest may be shaped by news and social media coverage that foregrounds emotionally charged and distressing events (e.g., violence, harm, confrontations with police, McLeod, 2007). Because diaspora members often rely on secondary, online accounts, they may be especially susceptible to such skewed inputs, projecting worst-case scenarios and overestimating on-the-ground suffering. Distance may further amplify this tendency: greater physical and psychological distance is associated with higher-level, more abstract construal, which can render protest experiences less nuanced and monolithic (Trope & Liberman, 2010). In line with this account, greater activism was associated with less overestimation of residents’ negative affect, plausibly because it likely increased access to firsthand accounts and, for some participants, involved direct exposure through protest participation. Still, residents showed little error in predicting diaspora affect across levels of activism, suggesting that affective misperceptions might be unique to diaspora members, perhaps reflecting distance from events on-the-ground specifically.
Residents’ errors in predicting how diaspora members identified protest behaviors can be understood through Vallacher and Wegner’s (1989) action identification model, which distinguishes higher-level construals that emphasize motives, meaning, and consequences from lower-level construals focused on mechanics and execution. From this perspective, residents’ underestimation of diaspora members’ action identification may reflect not only an attribution of “distance,” but a perceived mismatch between execution and purpose under local constraints. That is, skepticism that tactically oriented guidance from afar is calibrated to the goals, risks, and trade-offs faced by those protesting locally. This interpretation helps explain why the bias does not diminish substantially with residents’ activism: it may operate as a culturally available narrative about remote activism, shaping judgments beyond what greater activism can correct.
One possibility is that residents are simply differentiating themselves from diaspora members, attributing to them an emphasis on the immediate ‘how’—what to do next and how to do it—over the ‘why,’ an inference the task’s binary format may have encouraged. Alternatively, residents may genuinely interpret diaspora discourse as emphasizing surface features of actions over how actions advance broader goals under local constraints. Diaspora members’ mediated engagement may also contribute, insofar as the online ecology of remote participation can disproportionately amplify performance and execution over purpose. Consistent with this account, higher offline engagement was linked to smaller action-construal prediction errors.
The results point to perceptual gaps between residents and diaspora members. Differences in participation mode (street-level vs. mediated/online), rather than time devoted, likely contribute to diaspora members’ overestimation of residents’ distress and residents’ tendency to see diaspora activism as more focused on execution than on goals and strategy. The next section situates these findings in work on online protest ecosystems and on the psychological processes that sustain activism under risk.
Contributions to Understanding Distant Activism and Protest Participation
Online Ecosystems and Cross-Border Misperceptions
These findings underscore the dual nature of online activism. Social media can facilitate transnational protest by enabling information-sharing, mobilization, and motivational support (Jost et al., 2018; Zhuravskaya et al., 2020). At the same time, online environments can widen perception gaps—through selective exposure, misinformation, and strategic manipulation that shape how residents and diaspora members interpret each other’s roles and experiences (Cinelli et al., 2021; Vosoughi et al., 2018). Digitally mediated movements may also be vulnerable to coordinated disinformation campaigns engineered to distort narratives and sow division (Bessi & Ferrara, 2016). Moreover, digital activism can sometimes suppress rather than strengthen participation, contributing to slacktivism, social loafing, and state countermeasures that stifle dissent (Kwak et al., 2018; Tufekci, 2017; Wilkins et al., 2019).
As governments adapt their responses to online activism, their countermeasures increasingly extend beyond protests to target diaspora communities and activists. Through surveillance, infiltration, and disinformation campaigns, states seek to reshape the diaspora members’ perceptions of their host countries or alienate them from home activists (Chenoweth, 2022; Dukalskis et al., 2024; Earl et al., 2022; Waldinger & Shams, 2023). An analysis of WeChat, a prominent Chinese social media platform, found that state-sponsored content emphasized racism and hate crimes against Asians in the U.S. to discourage diaspora engagement in transnational activism (Audrye, 2024; Chester & Wong, 2025). Other strategies include discrediting activists, delegitimizing their motives, and disconnecting them from local movements (Greitens, 2024; Tsourapas, 2021). Thus, mediated engagement can expand transnational mobilization, but it can also foster systematic resident–diaspora misperceptions that diffuse protest momentum and create openings for state and counter-movement manipulation.
Emotions in Activism and Sustained Protest Participation
Across residence groups, greater activism predicted more positive and less negative affect in protest representations, despite the high-risk nature of the protests. This pattern is consistent with the idea that sustained engagement can be supported by psychological processes that help people maintain commitment even when immediate outcomes are uncertain. It also aligns with evidence that activism is associated with hedonic, eudaimonic, and social well-being (Klar & Kasser, 2009) and, in some cases, with improved mental health (Mužík et al., 2025), though this may vary by context and generation (Haidt, 2024).
A candidate mechanism for this effect might be efficacy and empowerment: taking action, even without guaranteed results, can increase perceived agency and sustain commitment (Peak & McGarty, 2025). Identification with a movement, emotional responses to injustice, and moral obligation are additional pathways that can sustain activism in high-risk settings (Ayanian et al., 2021; van Zomeren, 2013). For example, across protests in Russia, Ukraine, Turkey, and Hong Kong, Ayanian et al. (2021) found that participation intentions were strongly linked to beliefs that one’s contribution helps build a broader movement. Emotions also have distinct motivational signatures: shame and anger may promote collective action (Grigoryan et al., 2025), whereas fear can inhibit it (Miller et al., 2009). Because the affective items in this study tap motivational dynamics that shape how protests feel (e.g., hope, confidence, efficacy vs. fear, doubt), the more positive protest representations among more active participants likely reflect stronger identification and perceived agency, with reinforcement stemming less from outcomes than from participation itself.
Implications for Activists and Policymakers
Recognizing that residents and diaspora members can share broadly similar protest representations yet still misperceive each other has practical implications for activists and policymakers. Diaspora and online activists may overestimate on-the-ground distress, overlooking positive experiences that can sustain participation. While their roles in information-sharing and amplification are vital, communication with residents may be more effective when it also engages the movement’s overarching purposes and motivational themes, not only tactics and logistics. Residents’ perceptions of diaspora members may be better calibrated by recognizing that impressions of diaspora activism can reflect the content and constraints of mediated engagement (e.g., a tactical “how” emphasis) rather than a lack of alignment with the underlying purposes and meanings of specific protest actions.
For policymakers, the findings suggest that diaspora-based interpretations of events should be contextualized before informing analysis or decisions. Selective attention to distressing or provocative episodes and a focus on repression or violence can crowd out other features of collective action (e.g., solidarity, empowerment), distorting assessments of protest dynamics. At the same time, diaspora networks remain valuable channels for information exchange and cross-border coordination. Their effectiveness, however, may hinge on reducing the perception gaps that lead residents to see them as disconnected from on-the-ground realities.
Limitations and Future Directions
The data come from a single protest context, so generalizability should be tested in other movements. The sample was also likely self-selected and politically engaged, limiting representativeness and the ability to capture broader public attitudes. Methodologically, the protest-representation tasks relied on hypothetical scenarios rather than direct reports of lived protest episodes. Responses likely reflect a blend of episodic recall and constructed experience, especially among diaspora participants with uneven street-protest exposure. Likewise, the “answer as if” instructions in the Other condition may have elicited some self-projection (what respondents would feel if positioned as the other group) in addition to meta-perceptions about actual outgroup members. Accordingly, the observed misperceptions are best interpreted as systematic differences in cross-group beliefs and representations rather than precise measures of interpersonal accuracy. Future studies could refine prompts to better separate recalled experience, constructed scenarios, and true meta-perceptions, and more directly model distinct negative-emotion profiles (e.g., approach-oriented anger vs. withdrawal-oriented fear).
Resident–diaspora perception gaps may be a recurring challenge in digitally mediated movements. Diaspora networks can disseminate information and mobilize global attention online, yet mediated exposure may bias their assessments of local protest experience, and residents may interpret diaspora efforts as miscalibrated to local constraints and priorities. Future research should test whether similar judgment gaps emerge across movements and identify communication practices that reduce them. As collective action increasingly relies on transnational networks, narrowing these gaps may be crucial for sustaining trust, coordination, and cohesion.